Three AI trends to watch, according to Nobel-winning economist Daron Acemoglu, offer a nuanced perspective on the technology’s evolving impact, challenging prevalent narratives of an impending ‘AI jobs apocalypse’. Two years after his paper estimated only a small boost to US productivity and no obviating need for human work, Acemoglu’s measured take continues to spark debate, even as AI advancements prompt some previously skeptical economists to reconsider.
While public discourse, from political rallies to grocery store conversations, frequently conjures images of widespread AI-driven layoffs, data consistently supports Acemoglu’s initial predictions: AI has not yet significantly affected employment rates. However, the technology has advanced considerably since his cautious forecast. The Financial Standard spoke with Acemoglu to understand if recent developments have altered his thesis and what genuinely concerns him beyond the immediate threat of Artificial General Intelligence (AGI).
AI Agents: Augmentation vs. Replacement
One of the most significant technical leaps in AI since Acemoglu’s 2024 paper is the rise of agentic AI – tools capable of independent operation to complete given goals, moving beyond simple chatbots. Companies increasingly position these agents as a one-to-many replacement for human workers, promising efficiency and cost savings.
“I think that’s just a losing proposition,” Acemoglu states, suggesting agents are better viewed as tools to augment specific tasks within a job rather than being flexible enough to manage an entire role.
His reasoning stems from the multifaceted nature of human work. An X-ray technician, for instance, juggles approximately 30 diverse tasks, from patient histories to organizing mammogram archives. Humans fluidly switch between formats, databases, and working styles. Acemoglu questions how many individual tools or protocols an AI would require to replicate such versatility. The true impact of AI agents on jobs hinges on their ability to orchestrate tasks as seamlessly as humans. While AI companies are fiercely competing to demonstrate agents’ extended independent operation without errors, often exaggerating results, Acemoglu believes many jobs will remain immune to an AI takeover if agents cannot fluidly transition between diverse tasks.
The AI Industry’s Economic Hiring Spree
Beyond the well-documented scramble for AI researchers, Acemoglu points to a less-discussed trend: AI companies are rapidly building in-house economics teams. OpenAI, for example, hired Ronnie Chatterji as chief economist in 2024 and announced his collaboration with former Obama advisor Jason Furman to research AI and jobs. Anthropic has assembled a council of 10 leading economists, and Google DeepMind recently brought in Alex Imas as its “director of AGI economics.”
Acemoglu notes his colleagues are being recruited for these pivotal roles. This trend makes strategic sense, as AI companies are acutely aware of growing public skepticism surrounding AI, largely fueled by employment concerns. These firms have strong incentives to shape the economic narrative around their technology, as evidenced by OpenAI’s latest proposals for industrial policy in the intelligence age.
“What I hope we won’t get,” Acemoglu cautions, “is that they’re interested in economists just to further their viewpoints or further the hype.”
This tension defines the nascent field of “AI economics,” raising concerns that some of the most influential research on AI’s impact on work may increasingly originate from entities with the most to gain from favorable conclusions.
The Usability Gap: AI Apps and Adoption
While many interact with AI via user-friendly chatbots, Acemoglu urges a comparison to earlier tech transformations driven by software like PowerPoint or Word. These tools offered immediate, intuitive usability, allowing anyone to install and leverage them productively, leading to widespread adoption.
“We have not seen the development of apps based on AI that have the same usability,” he observes. Despite the ease of chatting with an AI model, the average worker often requires significant time and effort to achieve practical, productive use. This usability gap partially explains why AI has yet to demonstrate a seismic impact on the job market or the broader economy. Therefore, one of the key AI trends to watch for Acemoglu is the emergence of AI applications that significantly enhance ease of use and integration into daily workflows.
Acemoglu acknowledges the inherent uncertainty in the current AI landscape, where anecdotal evidence of a worsening job market for graduates coexists with a lack of measurable AI impact on productivity. The stark contrast between confident rhetoric and the underlying uncertainty remains the most defining characteristic of the AI economy today, underscoring the importance of observing these specific developments.
